A Content-Based Relevance Feedback Model for Product Review Retrieval
Weixin Tian, Zheng Sheng, Anhui Wang · 2010
Product review is a kind of useful information on the web. This paper describes a relevance feedback model based on modifying relation for that information retrieval, which utilizes the feedback information not only on the term frequency but also on the deep semantic structures. To calculate the feedback values based on semantic structures, a modifying relations knowledge base (MRKB) is used to measure the similarity between the term in relevant document and the term to be expanded. We propose a method to calculate and adjust the term weight. Experiment shows that our method got higher performance than the baseline when applying to the product review dataset.